How much would YOU pay for a Spin compiler?
mpark
Posts: 1,305
Would anyone pay anything for a Spin compiler that came with source code (C#)?
Yours Curiously,
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Michael Park
PS, BTW, and FYI:
To search the forum, use search.parallax.com (do not use the Search button).
Check out the Propeller Wiki: propeller.wikispaces.com/
Yours Curiously,
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Michael Park
PS, BTW, and FYI:
To search the forum, use search.parallax.com (do not use the Search button).
Check out the Propeller Wiki: propeller.wikispaces.com/
Comments
$50-$100 bounty on a .spin compiler that ran on the propeller.
OBC
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
New to the Propeller?
Getting started with a Propeller Protoboard?
Check out: Introduction to the Proboard & Propeller Cookbook 1.4
Updates to the Cookbook are now posted to: Propeller.warrantyvoid.us
Got an SD card connected? - PropDOS
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.
www.brilldea.com·- Prop Blade, LED Painter, RGB LEDs, uOLED-IOC, eProto fo SunSPOT, BitScope
www.sxmicro.com - a blog·exploring the SX micro
www.tdswieter.com
Probably not, but it's possible. Having source or not wouldn't swing the decision.
It would be a hard sell, but what could convince me is ...
* Command line compiler plus IDE ( with command line compiler not switching to windows to ask me a question or to tell me something )
* Runs under Win98SE with good speed on a 1GHz CPU ( .Net 2.0 is generally okay )
* Supports #Define ( single and multi-line ), #Include, #If-Else-EndIf etc
* Allows PASM EQU ( "Fred EQU $1FF", or "Fred ORG $1FF" ) and doesn't complain when the address is over $1EF
* Supports templates or some mechanism for assembling LMM code which is in a user defined format
* Generates a symbol table in text format
* Generates a disassembed bytecode listing so we can verify your code generation easily
* Allows the inclusion of a custom Spin Interpreter kernel ( that's easy, can provide details )
It's probably also worth checking out -
propeller.wikispaces.com/Propeller+Tool+-+Enhancement+Requests
Source code or no...
If it does not run on OSX Intel and PPC, Linux x86-32 and X86-64 and old versions of windows it just won't fill the gap any more than the current parallax compiler and a VM..
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Pull my finger!
@mpark
Do you mean a SPIN bytecode compiler or SPIN to ASM compiler?
Regards,
Coley
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PropGFX Forums - The home of the Hybrid Development System and PropGFX Lite
Not really so demanding. If you have the skills to write a compiler in Java or C# you could probably do it in C++ just as easily. Or just plain C.
Its very easy to write command line programs that are cross platform in C++ or C (Only has to read and write files after all).
So if your source is available it's can just be recompiled for any new platform.
This makes the need for all the bloat and hassle a Java or C# VM go away.
As for what I would pay, well, nothing. If it's tied to the Windows platform and C# you would probably have to pay me to take it away.
Why:
1) I already have a Spin compiler for free.
2) That compiler already causes me the pain of having to use Windows or an emulator/VM under my preferred OS.
3) I prefer languages to not use indentation for syntax.
4) Might be interested in something that compiled Spin to LMM and was compatible/linkable with ImageCraft C.
5) But then why not just use C ?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Q1) Would you like a Spin compiler ?
Yes.
Maybe - if it runs on my OS.
No.
Q2) Would you pay for it ?
Yes.
Maybe - Depends how much
Perhaps - If it's worth it.
No - I'll wait for the free Bit Torrent version.
No - I'm not interested.
For me (Q1) Yes, (Q2) No.
I think what is needed is a compiler written in C that can be ported to different platforms. Also it is hard to sell something 3rd party when the original one is free (even if it is closed source).
But to answer your question, no. I use Spin but don't like it one bit. Its the only other language besides Java that makes me like C.
However you need to consider that its a proprietary language tied to a one of kind processor. Your market is going to be small at best and you'll never recoup your time and money. Just ask ImageCraft if doing C for the Prop was a money maker.
IMO you might want to consider doing a BASIC compiler with Prop extensions(not Pbasic), that generates native code. It would be a great entry point for people to try out the Prop and not have to learn a rather odd and proprietary language in order to do it.
As far as source code goes, C or Pascal. I wouldn't touch anything that used Java or C#.
FWIW
Something along these lines?
Writing Object Table at 1 PUB AAA3(X1, X2) : Result | X3, X4[noparse][[/noparse]400], X5, X6 X1 := (X4) Addr : 0018: 70 - (X4) READ DBASE Long Variable offset 4 Addr : 0019: 65 - (X1) WRITE DBASE Long Variable offset 1 INA := 1 Addr : 001A: 36 - Constant 1 (00000001) Addr : 001B: 3F B2 - (INA) WRITE Long Register 2 Long[noparse][[/noparse]$12345678] += 2 + 2 Addr : 001D: 37 00 - Constant 2 (00000002) Addr : 001F: 37 00 - Constant 2 (00000002) Addr : 0021: EC - + Addr : 0022: 3B 12 34 56 78 - Constant $12345678 (12345678) Addr : 0027: 36 - Constant 1 (00000001) Addr : 0028: D2 4C - (Long) ASSIGN Absolute Long POP Address POP Index MATHOP + --INA Addr : 002A: 3F D2 36 - (INA) ASSIGN Long Register 2 --PRE INA++ Addr : 002D: 3F D2 2E - (INA) ASSIGN Long Register 2 POST++ INA~ Addr : 0030: 3F D2 18 - (INA) ASSIGN Long Register 2 POST~ INA~~ Addr : 0033: 3F D2 1C - (INA) ASSIGN Long Register 2 POST~~ IF 1 Addr : 0036: 36 - Constant 1 (00000001) Addr : 0037: 0A 05 - JZ Label0001 X1 := 1 Addr : 0039: 36 - Constant 1 (00000001) Addr : 003A: 65 - (X1) WRITE DBASE Long Variable offset 1 Addr : 003B: 04 80 52 - JMP Label0002 Addr : 003E: - Label0001 ELSEIF 2 Addr : 003E: 37 00 - Constant 2 (00000002) Addr : 0040: 0A 0D - JZ Label0003 IF 16 Addr : 0042: 37 03 - Constant 16 (00000010) Addr : 0044: 0A 03 - JZ Label0004 X1 := 16 Addr : 0046: 37 03 - Constant 16 (00000010) Addr : 0048: 65 - (X1) WRITE DBASE Long Variable offset 1 Addr : 0049: - Label0004 Addr : 0049: - Label0005 X1 := 2 Addr : 0049: 37 00 - Constant 2 (00000002) Addr : 004B: 65 - (X1) WRITE DBASE Long Variable offset 1 Addr : 004C: 04 80 41 - JMP Label0002 Addr : 004F: - Label0003 ELSEIFNOT 3 + (6 + 4 - -25 / 24) + (X4) Addr : 004F: 37 21 - Constant 3 (00000003) Addr : 0051: 38 06 - Constant 6 (00000006) Addr : 0053: 37 01 - Constant 4 (00000004) Addr : 0055: EC - + Addr : 0056: 38 18 E7 - Constant 25 (FFFFFFE8) Addr : 0059: 38 18 - Constant 24 (00000018) Addr : 005B: F6 - / Addr : 005C: ED - - Addr : 005D: EC - + Addr : 005E: 70 - (X4) READ DBASE Long Variable offset 4 Addr : 005F: EC - + Addr : 0060: 0B 04 - JNZ Label0006 X1 := X2 Addr : 0062: 68 - (X2) READ DBASE Long Variable offset 2 Addr : 0063: 65 - (X1) WRITE DBASE Long Variable offset 1 Addr : 0064: 04 2A - JMP Label0002 Addr : 0066: - Label0006 ELSEIF X1 //= 24 - X2 + (X1 + $321 / -24 + 6 - -X3) / X4[noparse][[/noparse]X5] + ^^ X6 Addr : 0066: 38 18 - Constant 24 (00000018) Addr : 0068: 68 - (X2) READ DBASE Long Variable offset 2 Addr : 0069: ED - - Addr : 006A: 64 - (X1) READ DBASE Long Variable offset 1 Addr : 006B: 39 03 21 - Constant $321 (00000321) Addr : 006E: 38 17 E7 - Constant 24 (FFFFFFE9) Addr : 0071: F6 - / Addr : 0072: EC - + Addr : 0073: 38 06 - Constant 6 (00000006) Addr : 0075: EC - + Addr : 0076: 6C - (X3) READ DBASE Long Variable offset 3 Addr : 0077: E6 - - Addr : 0078: ED - - Addr : 0079: CC 86 50 - (X5) READ DBASE Long Memory Address 0650 Addr : 007C: DC 10 - (X4) READ DBASE Long POP Index Memory Address 0010 Addr : 007E: F6 - / Addr : 007F: EC - + Addr : 0080: CC 86 54 - (X6) READ DBASE Long Memory Address 0654 Addr : 0083: F8 - ^^ Addr : 0084: EC - + Addr : 0085: 66 D7 - (X1) ASSIGN PUSH DBASE Long Variable offset 1 MATHOP // Addr : 0087: 0A 05 - JZ Label0007 X3 := 4 Addr : 0089: 37 01 - Constant 4 (00000004) Addr : 008B: 6D - (X3) WRITE DBASE Long Variable offset 3 Addr : 008C: 04 02 - JMP Label0002 Addr : 008E: - Label0007 X1 := X2 Addr : 008E: 68 - (X2) READ DBASE Long Variable offset 2 Addr : 008F: 65 - (X1) WRITE DBASE Long Variable offset 1 Addr : 0090: - Label0008 Addr : 0090: - Label0002 --X4[noparse][[/noparse]55] Addr : 0090: 38 37 - Constant 55 (00000037) Addr : 0092: DE 10 36 - (X4) ASSIGN DBASE Long POP Index Memory Address 0010 --PRE X5 := ~~X4[noparse][[/noparse]70] Addr : 0095: 38 46 - Constant 70 (00000046) Addr : 0097: DE 10 94 - (X4) ASSIGN PUSH DBASE Long POP Index Memory Address 0010 ~~PRE Addr : 009A: CD 86 50 - (X5) WRITE DBASE Long Memory Address 0650 X6 := 3 Addr : 009D: 37 21 - Constant 3 (00000003) Addr : 009F: CD 86 54 - (X6) WRITE DBASE Long Memory Address 0654 Addr : 00A2: 32 - Return▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Pull my finger!
Addr : 006E: 38 17 E7 - Constant 24 (FFFFFFE9)
Looks like a missing "-", and the FFFFFFE9 should be FFFFFFE8 - And as I'm trying to get LCC working that really isn't a complaint, just a comment !
Just for comparison, here's my disassembly of the X1 //= line at 0066 ...
I assume that listing was generated by a program. You are one step away from providing spin source/bytecode line translation. No? I guess accomodating everyone's style in the lookup could prove difficult.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
--Steve
Anything after the assembly code might well be slightly incorrect as I alter a lot of stuff in pre-processing. I've probably forgotten to change the descriptor on the token when I've done a unary negate. My benchmark is a bit for bit binary compatibility with the parallax compiler.. and that it does..
No, it's not finished..
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Pull my finger!
If you look closely, you will see the bytecode is identical.. It's my description that sucks.
What I've done is taken a constant and negated it. I've absorbed the unary '-' into the constants bytecode, so in the disassembly listing you wont see the '!', but it's there tacked on the end of the constant. Like I said, the stuff in the list you saw is kinda nasty internal debugging output rather than a decent list.. I'll need to fix that.. I'm pretty sure I know how to make it work effectively.
I've done a binary diff and the bytecode is identical, so I win there [noparse]:)[/noparse]
If you look at my disassembler output, you see pretty much what yours gives
(Ignore the addresses, I've added stuff to the top of the method since then)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Pull my finger!
PropII should be able to run FemtoBASIC at 10x the current speed with many
times more memory. You'll get your wish soon..
(At least by present standards.)
OBC
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
New to the Propeller?
Getting started with a Propeller Protoboard?
Check out: Introduction to the Proboard & Propeller Cookbook 1.4
Updates to the Cookbook are now posted to: Propeller.warrantyvoid.us
Got an SD card connected? - PropDOS
In terms of problems I would like to see solved... it could all be done in a pre-compiler or Chip is planning to do it anyway.
Here is a problem that is so mundane that I don't think Chip will ever waste his time on it, but someone should.
Let's say that I have a choice of graphics devices... LCDs, TV, uoleds, etc etc... and I want to make a product and when I get all done I change my mind... what do I have to do?
It certainly isn't go to my pre-compiler and change a check box is it?
And that is just one of the many things options that a designer has to consider.
The market is about a year away, you guys have plenty of time to do it. How about a committee?
Rich
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Pull my finger!
What you want is a Preprocessor. For that exists minimal 2 ready to use solutions:
My PreSpin: http://forums.parallax.com/showthread.php?p=655470
but you need version 0.2 from: www.insonix.ch/propeller/prespin.html
and Epmoyer's ScUMM: http://forums.parallax.com/showthread.php?p=654555
BTW: don't think you can make money with such Tools for the Propeller. PropTerminal, PreSpin and PASD have been downloaded hunderts of times, but not one donation, so far.
Andy